Tuesday, February 7, 2012

Very Unlikely Things Happen All the Time

On Sunday the NFC team won the opening coin toss at the Super Bowl for the 14th consecutive time. The odds against this are 8,192 to 1. So this qualifies as an event not likely to have happened by chance. In scientific terms, this is a 3.8 sigma result, and if you got one in your experiment you would be rushing to ready your paper for Science.

Most likely, though, this event means nothing. So many things happen every day that even highly unlikely ones show up in the sample. Somebody wins the lottery, even though the odds against anyone doing so are astronomical. I remembering reading an article by a risk assessment expert who said that if at the start of a year we made a list of a million things that had a 1 in a million chance of happening that year, at the end of the year we would ignore the 999,999 that didn't happen and insist that the one that did happen was much more likely than had been claimed. Even in scientific experiments we are generating so many results that very unlikely results happen every day.

Just because something is unlikely to have happened by chance doesn't mean it didn't happen by chance.

Then again, maybe the NFC has rigged the coin.

No comments: